Record and inspect MCP stdio sessions. A tracing and debugging tool for Model Context Protocol servers that captures request/response transcripts, detects latency regressions, and identifies tool contract changes (rug pulls).
MCPTracer is a tracing/debugging tool, not a production MCP server. Tool definitions are sparse and inferred from example demo files rather than explicit server registration. Only 2 of 8 tools have descriptions; most lack parameter documentation. Schemas are minimal (search has only 'query', send_email has 'to'/'body' but no descriptions). The 'echo' and 'mrtr-echo' tools appear to be test fixtures. Duplicate send_email entries (5 copies) suggest incomplete or malformed tool registration. No evidence of error handling, output schemas, or composition patterns. STDIO-only transport caps score at 50 maximum; actual definition quality is well below that.
Search the product catalog.
Send a plain-text email to the given recipient.
Send a plain-text email to the given recipient.
Send a plain-text email to the given recipient.
Send a plain-text email to the given recipient.
Send a plain-text email to the given recipient.
Tool definitions inferred from example files, not explicit server registration. Cannot verify actual tool implementation or schema completeness.
Duplicate send_email tool registered 5 times with identical schema. Indicates malformed tool registry or incomplete example cleanup.
Parameter descriptions missing for all tools. 'query', 'to', 'body', 'text' parameters lack context explaining what values are expected or how they are used.
No output schemas documented. LLMs cannot predict what search(), send_email(), or echo() return, forcing guesswork about downstream tool chaining.
Inferred effective spec: 2026-07-28+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 36 | 2026-07-28+ | v2 |
Test/demo tools (echo, mrtr-echo) exposed as production tools. No clear separation between tracing infrastructure and actual MCP server capabilities.